Tutorials

The general usage and workflow of lineshape_tools is demonstrated by the following tutorials. In general, there is a lot of overlap between steps in each tutorial. All of the tutorials culminate in evaluating the optical lineshape for a given case and differ in the way the dynamical matrix is generated.

• From a Foundation Model
  • Step 0: Identify and Calculate a Transition
  • Step 1: Compute the Dynamical Matrix
  • Step 2: Evaluate the Optical Lineshape
• From a Fine-Tuned Model
  • Step 0: Identify and Calculate a Transition
  • Step 1: (Optional) Generate Additional Training Data
  • Step 2: Collect Data
  • Step 3: Model Training
  • Step 4: Compute the Dynamical Matrix
  • Step 5: Evaluate the Optical Lineshape
• From Phonopy
  • Step 0: Identify and Calculate a Transition
  • Step 1: Obtain FORCE_CONSTANTS File Using Phonopy
  • Step 2: Evaluate the Optical Lineshape

When utilizing mace to generate the dynamical matrix, we recommend first using the foundation model and then fine-tuning to the relaxation dataset. (This is covered in the first and second tutorials.) These steps are fast and require no additional DFT calculations to be performed. One can then compare the results and assess whether further refinement is necessary before generating more data (the optional step in the second tutorial).

Python API

lineshape_tools was designed around the use of the command-line interface, but all of the inherent functionality is exposed in lineshape_tools.cli. We recommend first familiarizing yourself with the CLI, and then, depending what functionality you need, choosing the respective function in the API.